An evaluation of a pipeline planner for a filter based algorithm

Distributed query processing is one of the technical problems that need to be solved in Distributed Database Management Systems. Query Processing deals with designing algorithms that analyze queries and converts them into a series of data manipulation operations. The problem is how to decide on a strategy for executing each query over the network in the most cost effective way.;Through the past years, the research focus in distributed query processing has been on how to realize join operations with different operators such as Semi-join, Two-way Semijoin, and Pipeline N-Way joins. However, these operations will be executed sequentially, which may increase the data transfer cost.;A new algorithm, filter based pipeline N-way join algorithm, is presented to reduced data transfer cost. It makes use of filter concept and ensures the lower data access cost. This algorithm has three phases. Phase One: Use bloom filter to do forward semijoin and build tuple connectors. Phase Two: Do backward semijoin and build pipeline cache planner. Phase Three: Send Pipeline Cache Planner to query site.;The main goal for this new algorithm is to reduce data transfer cost while maintain low I/O cost as pipeline N-way join algorithm.